The purpose of this research is to specify and estimate an econometric model of the relationships among drinking alcohol, drunk driving, and drunk-driving countermeasures. Traffic accidents are one of the leading causes of death in the United States, and it is estimated that alcohol is involved in approximately one-half of all fatal traffic accidents. The proposed work will evaluate the effectiveness of recent policy initiatives following two approaches to reducing drunk driving: (i) alcohol control policies; and 1) drunk-driving deterrence. One equation to be estimated is an alcohol demand function, which shows drinking as a function of price, alcohol availability, health knowledge measures of the perceived costs of drunk driving, and individual characteristics. This specification will permit a test of whether laws that increase the certainty, severity, or swiftness of drunk-driving penalties act to deter the drinking that precedes drunk driving. The proposed work also will estimate what will be termed a demand function for drunk driving. Inclusion of the same measures of the perceived costs of drunk driving allows us to test whether drunk-driving countermeasures directly deter drunk driving. The specification will provide estimates of the total effect of the drunk-driving countermeasures on the amount of drunk driving, and the conditional effect on the amount of drunk driving controlling for the amount of drinking. The proposed study will use micro data from two cross-sectional surveys of individuals: the 1985 Health Interview Survey, and the Canadian National Survey on Drinking and Driving 1988. Additional information on state laws relating to drunkdriving countermeasures will be merged with the Health Interview Survey data. Responses to the Canadian survey will be used to create direct measures of individuals, perceptions of components of the costs of drunk driving, including the perceived probability of arrest and conviction for driving while impaired.